INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Circadian rhythms are universal biological phenomena found in eucaryotes and some procaryotes, and they are thought to be an adaptation to environmental cycles. A circadian clock governs the rhythms through physiological and endocrinological processes. The circadian clock keeps its period even when there are no environmental time cues. In addition, the clock's free-running period remains relatively constant with a change in temperature of 10°C, and the temperature quotient, Q₁₀, is approximately 1. The biochemical mechanisms underlying circadian rhythms are clearly distinct from biochemical reactions observed in other physiological and developmental events, because the Q₁₀ of those reactions is nearly 2 to 3.

Genetic studies using Drosophila mutants have facilitated our understanding of the molecular bases of the circadian clock. Five genes in Drosophila, period (per), timeless (tim), dClock (dClk), cycle (cyc), and double-time (dbt), have been identified as clock genes that contribute to a central oscillator mechanism (1, 31, 36, 40). The abundance of per and tim mRNAs and their protein levels fluctuate in a circadian manner (40), and dCLK and CYC are thought to form a heterodimer to act as transcriptional activators of per and tim (6). PER interacts with TIM (12), moves from the cytoplasm into the nucleus (5), and feeds back to repress the level of the per and tim transcripts (14, 16).

Although the molecular mechanism used to generate circadian fluctuation has been extensively studied, there are only a few molecular studies in Drosophila on the temperature compensation mechanism. At a behavioral level, per mutants affect not only period length but also temperature compensation; per^T and per^L mutants slightly shorten and lengthen their periods, respectively, as temperature increases (8, 18, 19). Several molecular studies suggested that temperature compensation is closely associated with PER. Huang et al. (15) reported that PER can undergo a temperature-independent intramolecular dimerization, while Gekakis et al. (12) showed that PER^L exhibits a temperature-dependent defect in binding to TIM although the molecular interaction between TIM and PER is temperature compensated. Moreover, an allele of the tim gene, tim^SL, can compensate for a temperature-dependent period lengthening of per^L (35). The length of the Thr-Gly repeat in PER is also reported to affect the temperature compensation (38).

We previously isolated ritsu (rit), a clock mutation on the second chromosome, from a natural population (26). We have now investigated features of rit and its interaction with per and tim at both the behavioral and molecular levels. rit mutants show abnormal temperature compensation of period and reduce PER and TIM levels. Molecular genetic analyses show that rit has a point mutation in the tim gene that leads to a single amino acid change, indicating rit is an allele of tim. Since an excess of the per gene dosage ameliorates the weak and delayed nuclear localization of PER as well as all other phenotypes of rit at both the molecular and behavioral levels, PER abundance in nuclei seems to be a key factor in the temperature compensation mechanism.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Stocks, locomotor rhythm recording, and mating procedures. Flies were kept under LD12:12 (12 h of light and 12 h of dark) at 24°C. Canton-S was used as the wild type. per; rit double mutants were synthesized by standard crosses. Flies were grown at 24°C. Locomotor activity was recorded as described elsewhere (25). The period of a locomotor rhythm was calculated by chi-square periodogram analysis (43). Mating for the recombination test between tim and rit (see Fig. 5A) was done as follows. To produce flies carrying both the rit mutation and the per-lacZ fusion gene on the second chromosome, we crossed rit; ry females to per-SG:3; ry males which carry the per-lacZ fusion gene on the second chromosome. These strains carried the ry mutation on the third chromosome, and this eye color should be rescued if a fly has the per-lacZ fusion gene. After two generations, we selected rit per-lacZ homozygous flies based on the ry⁺ eye color and the long-period rit phenotype. The four lines were selected as a rit per-lacZ strain. The rit rh-per strain was produced by standard mating procedures using SM1/Sco; TM3/Pr (21). +/w⁺Y; rit, C(1)DX, y w f/w⁺Y; rit, Dp(2;Y)odd^4.31; rit, Dp(2;Y)odd^2.31; rit and w; rit strains were produced by mating procedures described elsewhere (25), with minor changes. The recombination test was done by two different mating procedures. The rit/tim females were mated to tim⁰¹ males in one cross and mated to SM1/Pm males in the other crosses. SM1/Pm flies show a normal rhythmicity and are designated +/+ in Fig. 3A (right). In both cases, progenies from these crosses were then monitored for locomotor rhythms at 30°C. If recombination between tim and rit occurs, there should be rit⁺ tim⁺/tim⁰¹ progenies whose rhythm is normal in the former cross. In the latter cross, rit tim/SM1 or rit tim/Pm double mutants whose rhythm is abnormal would be obtained if recombination occurs.

RNase protection assay. Flies were entrained in LD12:12 for 5 days before they were collected. Total RNAs were extracted from 50 fly heads in 500 µl of extraction buffer (15 mM sodium acetate, 5 mM EDTA, 1% sodium dodecyl sulfate [SDS], 0.01% diethyl pyrocarbonate, 50 mM Tris [pH 9.0]). RNase-free DNase (Boehringer) was used to remove contaminated DNA. RNase protection assays were done as described elsewhere (14), with minor modifications. We used the per5 and TIMAX1 probes and ribosomal protein 49 (rp49) as a control. The per5 probe is a genomic fragment of the per gene containing about 210 bp (bp 5849 to 6060) of the per exon 5. TIMAX1 is a cDNA fragment of the tim gene from bp 4963 to 5192 (39). We found that the abundance of rp49 in samples obtained at 30°C was half as much as that at 24°C, although the reason was unknown. Therefore, each measurement was normalized by the value of the peak level for wild-type flies at each temperature.

Immunoblot analyses. Protein extracts were made from 50 fly heads for each time point as described by Edery et al. (7), with minor modifications. Each sample was homogenized in 15 µl of ice-cold extraction buffer, and the tip of the homogenizing pestle was rinsed twice with another 15 µl of extraction buffer. Amounts of proteins in a total of 45 µl of extraction buffer were measured by the Bradford protein assay system (Bio-Rad). After the measurement, extraction buffer was added to bring the protein concentration to 5 µg/µl in each sample. Five-microliter aliquots of 3× SDS sample buffer were added to 10-µl samples and boiled for 5 min. Supernatants were loaded on SDS-5% polyacrylamide gels. Western blotting was done as described previously (7), with minor modifications. We used anti-TIM antibody (donated by J. Blau) diluted 1:5,000. Horseradish peroxidase (HRP)-conjugated anti-rat immunoglobulin G (IgG) antibody (Cappel) was used as a secondary antibody, diluted 1:3,000. For quantitating chemiluminescence (SuperSignal CL-HRP substrate system; Pierce), an exposure on X-ray film (Fuji film) was digitally imaged by Densito Graph (ATTO) and quantified by NIH Image software. After the exposure, the membrane used was incubated for 3 h in the substrate buffer to eliminate HRP activity. Then the membrane was used to detect PER abundance as follows. We used anti-PER antibody (donated by R. Stanewsky) diluted 1:10,000. HRP-conjugated anti-rabbit IgG antibody (Cappel) was used as a secondary antibody, diluted 1:3,000. Exposure and quantification were done as described above.

Histology. Expression of the per gene in rit flies was assayed histologically by using the per-lacZ fusion gene as a reporter. Flies at ZT18 (zeitgebertime 18 h) and ZT6 were frozen in liquid nitrogen and mounted into O.C.T. compound (Tissue-Tek). Sections of 10 µm were cut and stained with 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal). The staining procedure was done as described by Liu et al. (22). Head sections of wild-type and rit flies were embedded side by side on the same slide to compare the staining profiles of the two strains. Photographs were taken with a Zeiss AxioPhot microscope. For fluorescent immunostaining with anti-PER antibody, the white-eyed strain was used as the wild type to eliminate background fluorescence of eye pigment. For the same purpose, the w; rit double mutant was used. Sections (14 µm) of the two strains were embedded side by side on the same slide to facilitate comparison of the strengths of signals. Anti-PER antibody was applied at a dilution of 1:15,000. Anti-rabbit IgG conjugated to peroxidase-labeled dextran polymer (EnVision +; Dako) was applied as a secondary antibody. Signals were enhanced with fluorescein isothiocyanate (FITC)-Tyramide (NEN). The Tyramide signal amplification reaction was usually done for 7.5 min; to obtain a maximum sensitivity, it was extended to 15 min. Counterstaining of nuclei was done with propidium iodide (1 µg/µl; Sigma) after RNase (10 µg/µl; Boehringer) treatment for 30 min. The double-staining images were visualized with a Zeiss LSM410 confocal laser scan microscope equipped with a krypton-argon laser.

Reverse transcription-PCR and cDNA sequence. Total RNA from 50 heads obtained from rit flies at ZT18 at 24°C were reverse transcribed by using a Ready-To-Go T-primed first-strand kit (Pharmacia). Using four tim-specific primer sets (tim237-258 plus tim1245-1226, tim914-933 plus tim1835-1816, tim1813-1834 plus tim3404-3383, and tim3122-3142 plus tim4474-4453; numbers are based on the nucleotide position of the tim cDNA [28]), fragments were amplified and cloned into the pCRII vector (Invitrogen). Amplified fragments too large to be sequenced were digested by restriction enzymes and subcloned. Both strands of each clone were sequenced at least twice with an ALFred DNA sequencer (Pharmacia). Experiments were repeated at least twice to avoid PCR errors. Fragments from tim bp 3122 to 4553 amplified from the wild-type or rit strain, a region which includes the nucleotide change from CCG to GCG, were digested by EcoRI at 37°C for 3 h. The digestion occurred in the wild-type fragment but not in the rit fragment.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

rit alters free-running periods in a temperature-dependent manner. Locomotor activity rhythms of individual flies were recorded under LD12:12 for 3 days followed by constant darkness (DD). The rit strain was entrained to LD12:12 at 24°C and showed a lengthened circadian rhythm under DD; its period was about 2 h longer than that of the wild type (Fig. 1A and C). We then measured the period of rit flies at different temperatures (Fig. 1A). When the temperature was lower than 24°C, the period of rit flies was only slightly lengthened, with a Q₁₀ of 0.93, which was comparable to that of wild-type flies (Q₁₀ = 1). When the temperature was above 24°C, the period of rit flies lengthened remarkably to about 10 h longer than that of wild-type flies at 30°C (Fig. 1A). Thus, above 24°C, the Q₁₀ of rit flies was 0.62, which is significantly different from the wild-type value of 1. This phenotype is recessive since the period of heterozygous rit/+ flies were well temperature compensated (Fig. 1C).

View larger version (55K): [in this window] [in a new window]   FIG. 1.   rit lengthens circadian periods as temperature increases. (A and B) Locomotor activity records at various temperatures for rit (A) and perL; rit (B) flies. Flies were held in LD12:12 for the first 3 days and then kept in DD. Light regimens (white bars, light; black bars, dark) are indicated above the tetraplotted (4 days) actograms. (C) Changes in period at various temperatures. Mean values at 20, 28, and 30°C were calculated to combine data at 19 and 20, 27 and 28, and 29 and 30°C, respectively. , strains carrying no mutation on the second chromosome (Canton-S, perL, and perS); , strains carrying the rit mutation (rit, perL; rit and perS; rit). , heterozygous rit/+ strain. Vertical bars show standard errors of the means. Numbers beside symbols represent the numbers of rhythmic flies. Q10 values are separately represented when the changes in period are different between the temperature ranges below and above 24°C.

Genetic interaction between rit and per. rit strongly interacts with per^L with respect to period lengthening. The double mutant of rit (25.5 h at 24°C) with per^L (28.5 h at 24°C) showed a period of 32.9 h at 24°C (Fig. 1B). This is ~2 h longer than the value expected (31 h) on the basis of an additive effect of the two mutations. At 27°C, per^L; rit flies showed an extremely long period of 44.4 h (Fig. 1B). Although these periods are extraordinary long, the rhythmicity itself was still clear and stable, with a punctual onset and offset of the active phase. While only 2 of 25 flies were arrhythmic at 27°C, most rit flies became arrhythmic at 30°C (Table 1). One fly that was rhythmic at 30°C revealed an extremely long period of 50.4 h (Fig. 1B). Even in this case, we observed a stable rhythmicity. The Q₁₀ of the double mutant was drastically lower (Q₁₀ = 0.46) than that of the per^L mutant (Q₁₀ = 0.88) above 24°C.

rit induces arrhythmicity at higher temperature. We observed that a majority of rit flies become arrhythmic at temperatures higher than 24°C (Table 1). When the temperature was 24°C, flies were quite rhythmic in all strains used in this study. At the lowest and highest temperatures tested (16 and 30°C), a portion of wild-type flies and of per mutants showed arrhythmicity (Table 1). This is because overall locomotor activity tends to be reduced at 16°C (data not shown). At high temperatures (30°C), some flies were statistically arrhythmic, although a weak but stable rhythmicity could be observed by eye in all plots.

An excess of per gene dosage ameliorates the rit phenotype. rit males and females, carrying a per⁺ gene translocation on the Y chromosome (w⁺Y), were produced to test if an additional per⁺ dosage affects periodicity in rit. In +/w⁺Y; rit male flies, the per gene dosage is double that of a wild-type male, while attached-X [C(1)DX]/w⁺Y; rit females have 1.5 times more per gene dosage than wild-type males because per is on the X chromosome (3). A per gene dosage of 2 completely rescued the rit phenotype behaviorally; its period was normal and temperature compensated (Fig. 2A). When the per gene dosage was 1.5 in C(1)DX/w⁺Y; rit flies, the rit phenotype was rescued at lower temperature but incompletely rescued at higher temperature (Fig. 2A).

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Duplication of the per+ and tim+ genes restores the extra long period in rit flies. (A) The rit phenotype is restored by the per+ gene translocation in a dosage-dependent manner. +/w+Y; rit is the rit male carrying the per+ gene translocation on the Y chromosome; C(1)DX represents females carrying the attached X chromosome. (B) Complementation test with tim01. tim01 does not complement rit, and the tim+ gene translocation restores the rit phenotype. The per+ gene translocation also restores the phenotype shown in rit/tim flies. (C) The rit phenotype is restored by the tim duplication. Dp(2;Y)odd4.13 and Dp(2;Y)odd2.31 carry the tim+ gene translocation on the Y chromosome. Numbers beside symbols represent the numbers of flies showing rhythmicity at each temperature.

Complementation and recombination tests with tim. The rit locus was roughly mapped near the cl gene (26). Since this position is near the tim locus, rit could be an allele of tim. To determine if this is the case, we tested whether rit could complement tim⁰¹ (a null allele of tim [28]). Locomotor activity rhythms of tim⁰¹/rit flies were recorded at various temperatures from 19 to 30°C (Fig. 2B). rit/tim⁰¹ heterozygotes phenocopied the rit phenotype. The period in these heterozygotes at each temperature was, however, about 1 h shorter than the period in rit homozygotes (t test, P < 0.05). We obtained a similar result in assays using Df(2L)tim⁰² (data not shown), which deletes the entire tim gene (28). Since the temperature dependency of period in rit/tim heterozygotes phenocopied rit homozygotes and the phenotype of rit is rescued by duplication of the per locus as described above, we expected that the phenotype in rit/tim heterozygotes would also be rescued by the per duplication. Male flies having a per gene dosage of 2 showed normal periods at various temperatures (Fig. 2B).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   rit is a novel allele of tim. (A) Mating schemes to test whether recombination occurs between rit and tim01. All genotypes and their expected period in the second generation, assuming that recombination occurs, are listed. Rhythmicities were judged by chi-square periodogram analysis (43) ranging from 19 to 29 h. The arrhythmic category includes flies showing an extra long period over the circadian range. See Results for details. (B) Schematic representation of the coding region in the timrit cDNA. The coding sequence is indicated by a box. The arrow lines indicate PCR fragments amplified for sequencing. Amino acid numbering is as specified by Myers et al. (29); domains indicated by closed boxes are based on the study by Saez and Young (37). Met, translation start; NLS, nuclear localization signal; CLD, cytoplasmic localization domain.

rit encodes an amino acid substitution in TIM. The coding region of tim cDNA in rit flies was amplified by reverse transcription-PCR. The fragments amplified were sequenced and compared to the tim⁺ cDNA previously described (28, 29). To avoid PCR errors, experiments were independently repeated twice and sequence data were confirmed for each repeat. There were 16 mutations between rit and tim⁺ at the nucleotide level. Among these, 14 were silent mutations that do not cause a change in the amino acid sequence. We found a single-base deletion at position 294 (numbering according to references 28 and 29) in the noncoding region (Fig. 3B). This deletion has been described as a mutation which has no effect on circadian rhythm in three Drosophila species, including Drosophila melanogaster (33). There is a missense mutation which produced an amino acid substitution at bp 3492 from the start point of the tim cDNA (Fig. 3B). At this point, the nucleotide change from CCG to GCG yields an amino acid change from proline to alanine (Fig. 3B) at amino acid 1093 in TIM protein (numbering according to reference 29). Since there is an EcoRI site at this position (Fig. 3B), we confirmed that this mutation abolishes the restriction of the tim^rit cDNA by EcoRI at this region (data not shown). Taken together, the results led us to conclude that rit is an allele of tim. Hereafter, we use rit as an allele name and tim^rit to designate a mutation.

rit lowered expression and protein abundances in tim and per mRNAs. We measured per and tim mRNA cycling in fly heads by RNase protection assay. Flies entrained in LD12:12 were collected every 4 h. RNA abundance was normalized by the peak value (at ZT13) of the wild-type per and tim mRNAs at 24°C (Fig. 4A). Wild-type flies at 24°C exhibited robust mRNA cyclings with the peak at ZT13 to ZT17 and a trough at ZT1, meaning that these mRNA cyclings are in phase. The tim mRNA also cycled in rit flies at 24°C but with a delayed peak. The levels of tim mRNA in rit flies were not significantly different from those in the wild type at each time, while the peak value of per mRNA in rit flies was reduced to about 70% (t test, P < 0.05) (Fig. 4B). At 30°C, the peak of per and tim mRNA cyclings in the wild type was delayed compared to that at 24°C (Fig. 4C). The shape of tim mRNA cycling in rit flies at 30°C was quite similar to that at 24°C (Fig. 4D). The abundance of per mRNA at the peak (ZT17) decreased to 60% (t test, P < 0.05). Thus, the amplitude of per mRNA cycling becomes smaller as temperature increases, while tim mRNA is not affected in rit.

View larger version (24K): [in this window] [in a new window]   FIG. 4.   Daily patterns of per and tim mRNA cyclings under LD. RNase protection assays were performed on total RNAs from wild-type and rit flies entrained in LD12:12 at 24°C (A and B) and at 30°C (C and D). See Materials and Methods for details. Values at each point are means of three to eight experiments. Vertical bars show standard errors of the means. Light regimens are indicated (white bars, light; black bars, dark). We defined the lights-on point as ZT0 and the lights-off point as ZT12. Asterisks indicate that the mean value for rit flies is significantly different from that for wild-type flies (t test, P < 0.05).

View larger version (38K): [in this window] [in a new window]   FIG. 5.   PER and TIM abundance in wild-type and rit flies at various temperatures. Adult head homogenates were obtained from flies entrained at 24, 27, and 30°C and subjected to Western blot analysis using anti-TIM antibody followed by anti-PER antibody as described in Materials and Methods. TIM and PER bands in panel A were quantified by densitometry. Measurements obtained at each point were normalized by the maximum value for wild-type flies under each condition. (B) Mean abundances of TIM and PER. Values at each point were means of three to nine experiments. Vertical bars show standard errors of the means. Asterisks indicate that the mean value for rit flies is significantly different from that for wild-type flies (t test, P < 0.05). Data for rit flies obtained at 30°C were classified into two groups (bottom row in panel B; see also Results for details). (C and D) Under constant darkness at 30°C, no or very weak cycling is indicated by the data from 8 h of sampling in both proteins. Experiments were independently done two and three times for TIM and PER, respectively (D). Light regimens are indicated (white symbols, light; black symbols, dark).

rit affects per mRNA abundance at a posttranscriptional level. rit appears to lower per mRNA abundance and amplitude of PER protein cycling, while tim mRNA abundance is not affected but TIM protein abundance decreases. One possible reason for this finding is that rit primarily affects the per mRNA transcription level; another possibility is that rit reduces PER abundance, which in turn may decrease per mRNA abundance through the per feedback loop. To solve this problem, we produced a rit; rh-per strain and measured its PER abundance. In rh-per flies, per is strongly driven by rhodopsin1 promoter in eyes independent of the innate per gene expression (45). If PER is reduced in rit; rh-per flies, rit should affect PER at a posttranscriptional level. The PER level in the strain carrying rh-per was about 80% lower in rit⁺ flies than in rit flies. This is obvious at 30°C (Fig. 6A). This result suggests that rit affects the per feedback loop at a posttranscriptional level. PER seemed to be hypophosphorylated in the rit background at 30°C when the mobility shifts were examined by side-by-side comparisons at ZT2 (Fig. 6A).

View larger version (32K): [in this window] [in a new window]   FIG. 6.   rit affects PER abundance at a posttranscriptional level. (A) Constitutive expression of PER, using the rh1-per fusion gene. Flies were collected at ZT2 except for one wild-type strain collected at ZT18 as a control. PER bands detected by Western blot analysis using anti-PER antibody were quantified and were normalized by the PER level of the wild type. +, wild-type (Canton-S); rit/+, heterozygous rit; rh-per, protein extracts were obtained from flies carrying the rh-per construct. Lysates were obtained from the following strains: wild type as a control (lane 1), wild type (lanes 2 and 7), rit (lanes 3 and 8), rh-per (lanes 4 and 9), rit; rh-per (lanes 5 and 10), and rit/+; rh-per (lanes 6 and 12). Values at each point were means of four experiments. Asterisks indicate that the mean value for rit; rh-per flies is significantly different from that for rh-per flies at the same temperature (t test, P < 0.05). Vertical bars show standard errors of the means. (B) Daily fluctuation of PER and TIM in rit flies with an excess per gene dosage. Adult head homogenates were obtained from flies entrained at 30°C. Values at each point were normalized by the maximum value for the wild type. Values at each point were means of four experiments. Vertical bars show standard errors of the means. Light regimens are indicated (white bars, light; black bars, dark).

Levels of PER translocated into the nucleus are lowered in a rit background. The spatial pattern of per expression can be monitored by lacZ expression in transformant flies carrying a per-lacZ fusion gene (22). This fusion gene contains one-half of the per coding region, ca. 4 kb of 5' flanking region, and the entire coding region of the lacZ gene derived from Escherichia coli. per-lacZ and rit per-lacZ flies were entrained under LD12:12 at 24 or 30°C for 3 to 5 days. Sections of wild-type and rit flies were incubated with X-Gal for 2 h at 37°C. In both strains at 24°C (Fig. 7A and B), there were lacZ-positive cells in optic lobes (lamina and medulla) and the central brain. The expression pattern was principally coincident with the previous study (22). Nuclei in photoreceptor cells were stained strongly in wild-type but only weakly in rit flies at 24°C.

View larger version (135K): [in this window] [in a new window]   FIG. 7.   Nuclear localization of PER is impaired in rit flies. Horizontal section of fly heads stained with X-Gal (A to D and E). The spatial pattern of per gene expression was monitored by using the per-lacZ fusion gene. Horizontal head sections in wild-type and rit flies kept at 24°C (A and B) or 30°C (C and D) were stained for 2 h at 37°C. per was expressed in retina (ret), optic lobes (lamina [la], medulla [me], and lobula [lo]), and central brain (br) in wild-type flies (A and C). Nuclei in photoreceptor cells are clearly stained (arrowheads) in wild-type but not rit flies at both 24 and 30°C. Staining of LNs is indicated by black arrows. The expression level in the PER--Gal fusion protein in rit flies is lower than that in wild-type flies. Such a weak PER localization in nuclei was rescued by the excess of per+ gene dosage in +/w+Y; rit even at 30°C (E). The scale bar represents 80 µm.

View larger version (81K): [in this window] [in a new window]   FIG. 8.   PER localization in lateral neurons at 30°C. Horizontal sections of fly heads obtained at ZT19, -21, and -23.5 were stained with anti-PER antibody coupled to FITC (green color). Counterstaining of nuclei was done with propidium iodide (red color) after RNase treatment. When PER staining overlaps nuclear staining, yellow color is observed. For symbols, see the legend to Fig. 7. In wild-type flies (A to C), PER signals are comparable to those seen with X-Gal staining in Fig. 7C. LNs at a magnification of ×8 are represented in panels D to F. PER signal is observed in the cytoplasm at ZT19 and in nuclei at both ZT21 and ZT23.5. In rit flies (G to I), staining similar to that in wild-type flies can be observed, although the PER signal is weaker especially in the LNs (J to L). Thus, images in which PER signals were amplified are illustrated (M to O). PER is in the cytoplasm at ZT19 (J and M). At ZT21, there are three patterns of PER staining surrounding a nucleus (center and right upper corner in panel N), overlapping in a nucleus of the center of a broad PER staining area (middle left in panel N), and just overlapping in nuclei (inset in panel N). At ZT23.5, PER enters nuclei (O) or stays in the cytoplasm (inset in panel O).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

rit, a new allele of tim, lengthens or abolishes circadian locomotor rhythms in a temperature-dependent manner. Thus, tim^rit appears to affect a temperature compensation of period which is one of the most important features of circadian rhythms. However, rit can be regarded as a temperature-sensitive (ts) mutation lengthening a circadian period depending on temperature. Several studies have been performed to determine the mechanism of temperature compensation in Drosophila (12, 15, 38; reviewed in reference 13), but the molecular mechanism has not been defined. Therefore, at present we cannot distinguish a ts mutation of period from a temperature compensation mutation. However, the following results led us to consider how rit affects the circadian rhythm. A tim duplication rescues the abnormal phenotype of temperature compensation in the circadian period of rit, while the period of rit flies having the duplication of tim was still 2 h longer than in wild-type flies. In addition, rit/+ heterozygotes showed a period 1 h longer at all temperature ranges tested. These data mean that the tim^rit mutation is recessive for the temperature-dependent lengthening of period but semidominant for the temperature-independent lengthening of period. Thus, the tim⁺ gene enables the circadian system not only to maintain circadian period but also to compensate the period against temperature such that a half dosage of tim⁺ is enough for function. Our point that tim contributes to the temperature compensation mechanism as well as the maintenance of period is supported by studies of another tim allele, tim^SL. This mutant restores temperature compensation (or a ts effect of lengthening period) to per^L flies (35). Since tim^SL in the per⁺ background shows a nearly normal period at all temperature ranges tested (35), tim^SL may affect temperature compensation rather than keep circadian period constant. In summary, we conclude that rit is a mutation of temperature compensation rather than a ts mutation of period.

The mutation site of tim^rit, where the amino acid substitution occurs, is separated from that of tim^SL with respect to the primary structure of TIM. The mutation site of tim^rit is outside neither the nuclear localization domain nor the dimerization domain with PER (37). Given the behavior phenotype, the extra long locomotor period of rit flies at higher temperatures is similar to that reported for TIM1 transformant flies which lack a 32-amino-acid region in tim cDNA (30). Since our sequence data indicate that this region of tim cDNA is intact in rit, the function of the domain where the tim^rit mutation occurs is still unknown. Our preliminary computer simulation indicates that the tim^rit mutation may lead to a local conformation change in secondary structure of TIM from coil to  sheet at the tim^rit mutation site (24a). Through such a change in conformation, the tim^rit mutation may alter a protein-protein interaction between TIM and PER. Since PER abundance is lowered in rit flies and the excess of PER rescues the rit phenotype, it is certain that the tim^rit mutation alters circadian rhythm by affecting PER abundance.

The peak levels of PER and TIM are lowered at 30°C when rit flies shows arrhythmicity or an extra long period of locomotor activity rhythm. Thus, lower levels and amplitudes of PER and TIM cycling correlate with behavioral phenotypes. Since TIM not only acts as a nuclear transporter of PER (44) but also influences PER stability (16, 32), the low level of PER in rit flies is probably caused by the low level of TIM^rit. One question arising here is why PER and TIM show rhythmicity in abundance under LD but are basically arrhythmic under DD at 30°C. Null mutants of per and tim genes are arrhythmic at the mRNA and protein levels under both LD and DD (14, 16, 39). One possibility is that the rhythmicities of PER and TIM under LD at 30°C are partially driven by the light-dark cycle itself. Flies kept in LL phenocopy the tim⁰¹ mutant with regard to PER expression (32) because the abundance of TIM is rapidly reduced by light (16, 20, 27, 46). Since we confirmed that the level of TIM^rit is reduced under constant-light conditions (25a), the response of TIM to light seems not to be affected by the tim^rit mutation. Another possibility is that rit affects the mode of PER, especially its accumulation in nuclei through altering PER phosphorylation, because we showed that PER is hypophosphorylated in rit at 30°C. Interestingly, an amorphic allele of the dbt gene, dbt^P, shows the rit phenotype that mRNAs and protein levels of per and tim fluctuate in a nearly normal manner in LD but not in DD (31). DBT is a protein closely related to human casein kinase I (17), phosphorylating PER, and this phosphorylation is a key step of circadian fluctuation of PER abundance (31). The dbt locus on the third chromosome is intact in rit flies because the chromosome was changed to wild type in rit. Although little is known about the molecular link between tim and dbt except that the level of tim mRNA decreased prior to reduction of TIM abundance in dbt^P under DD (31) and that PER phosphorylation is absent in tim⁰¹ (31), our result suggests that TIM and DBT likely cooperate to phosphorylate PER.

If the sole function of TIM is to stabilize PER and to transport PER into the nuclei, our results for tim^rit cannot be explained with certainty. For example, if we think that arrhythmicity at higher temperature in rit flies is caused by a defect in the interaction between TIM^rit and PER, we cannot explain how the additional per⁺ dosage resulted in the complete recovery of arrhythmicity, except by assuming that PER alone could enter nuclei. Also it is hard to explain why PER and TIM levels in these flies increased even more than in the wild-type flies. These considerations lead us to assume that TIM has some additional role other than binding to PER. So and Rosbash (42) suggested that a posttranscriptional regulation step is involved in per mRNA oscillations. Available data do not support the view that TIM can bind to RNA, but one attractive hypothesis is that TIM is involved in the stability of per mRNA. Then TIM^rit may destabilize per mRNA in a temperature-dependent manner, thus lowering the PER level.

The effect of rit on per and tim transcriptional levels appears to be more severe in per than in tim. The peak level of per mRNA in rit flies decreases to 60 to 70% of that in wild-type flies at both 24 and 30°C. Since the tim^rit mutation probably affects the per feedback loop at a posttranscriptional level, the alteration of per mRNA levels should be induced through the per feedback loop. Recently, two Drosophila genes, Jrk (dClk) and cyc (dBMAL), were identified as transcriptional activators of per and tim genes (1, 6, 36). These genes are intact in rit because they are mapped on the third chromosome, which was changed to the wild-type chromosome in the rit strain. It is believed that per and tim transcriptions are regulated in a similar way through the regulatory element called the E box encompassing the per and tim promoters (6). However, our present results suggest that there are separate transcriptional regulations for per and for tim. Thus, we suspect that TIM^rit alters per transcription through interacting with Jrk, cyc, or other transcription factors which have not been identified, although the mode of protein-protein interaction among products of those clock genes is still unknown. Another possibility is that TIM^rit destabilizes the per mRNA through posttranscriptional regulation. Posttranscriptional regulation in per mRNA was reported by So and Rosbash (42), although they provided no evidence that TIM influences such regulation either in a direct or an indirect way.

We showed here that the excess of per⁺ gene dosage ameliorates the temperature dependency not only of period in a behavior rhythm but also of abundances of PER and TIM and the amplitudes of their cyclings. This finding suggests that PER abundance is an important factor in the temperature compensation mechanism. However, because the period of per/+ flies, in which per⁺ gene dosage should be half as much as in wild-type flies, is well temperature compensated (24a), it would be hard to say that the reduced PER level alone in rit directly reflects its abnormal temperature compensation of period. As PER acts in the nucleus rather than in the cytoplasm, it is essential to compare the amounts of PER in nuclei between wild-type and rit flies. Our histological study using anti-PER antibody and the per-lacZ reporter gene showed that the abundance of PER expressed in LNs was lowered in rit flies and the number of LNs in which PER enters varies among rit flies at 30°C. The nuclear entry of PER probably delays in its timing and/or does not occur in a part of LNs in rit flies. This phenotype can be rescued in +/w⁺Y; rit. Taken together, the key feature of the temperature compensation of period may be the amount of PER transported in nuclei by TIM. Furthermore, because the nuclear localization of PER and protein cyclings of PER and TIM are rescued only by the increasing per⁺ gene dosage at 30°C, the ability of the nuclear transportation is virtually normal in TIM^rit. The tim^rit mutation may first affect the interaction between PER and TIM and then disrupt their nuclear localization. Previous reports indicated that the per gene dosage negatively correlates with period length (4, 41). Additionally, Curtin et al. (5) reported that PER accumulates in the cytoplasm for several hours before entering nuclei in LNs and that the nuclear entry of PER is temperature sensitive in the per^L mutant whose mutation occurs in the PAS domain, where PER interacts with TIM. It could be concluded that the temperature-dependent lengthening or abolishment of period by tim^rit is presumably induced as follows: tim^rit affects the interaction between TIM^rit and PER, thereby decreasing the levels of PER and disrupting its nuclear localization. The low abundance of PER in nuclei makes it possible to interpret the long period of rit mutants based on the negative correlation with a gene dosage of per. Arrhythmicity observed in locomotor rhythm under DD might be induced by no or very little entering of PER into nuclei. This result is inconsistent with the fact that circadian locomotor rhythmicity can be rescued only if a very few cells in the restricted nervous system of a brain express the per gene (9, 10), because most of the rit flies lost the circadian rhythmicity of their locomotor activities while half of the nuclei in LNs seemed to be stained even at 30°C in many rit flies. One possible explanation is that the long or arrhythmic phenotype is caused by weak coupling among circadian oscillators in LNs.

In Neurospora, frq-9, an allele of the frequency gene, was isolated as a temperature compensation mutant whose Q₁₀ is about 2 (24). At a molecular level, it was reported that two initiation codons of the frq gene are selectively used at different temperatures (11), and a temperature-dependent threshold level of FRQ is required to establish the feedback loop comprising the oscillator (23). Although the possibility that alternative methionines are used as an initiation site in the tim gene has been proposed (29, 33), such regulation has not yet been reported in Drosophila clock genes. However, it is significant that there is a critical temperature at which the lengthening effect is much more severe in rit. The effect is not so drastic below 24°C, but a drastic change in period is observed above 24°C (Fig. 1C). Furthermore, half a gene dose of per⁺ was enough to rescue the rit phenotype at low temperatures. It is thus possible that there is a threshold level of PER required to establish the clock mechanism in a temperature-dependent manner.

One further issue to examine is whether transformant flies carrying the tim^rit point mutation mimic the rit phenotype in the tim⁰¹ background. Such analyses have been done in studies on the per^S mutation site (2, 34). Another issue is whether the PER-TIM interaction is affected in tim^rit as a function of temperature. We believe that these studies should give insights into the still mysterious molecular mechanism of temperature compensation of circadian rhythms.